on new desktops, far outpacing 40- and 80-MB flash-memory cards.
Under the covers, though, flash memory has become big in products that need a small amount of nonvolatile storage for programs and data. According to In-Stat, a Scottsdale, Arizona, market research firm, only about 25 percent of flash-memory products went into compute
rs in 1995, making computers the number-two users, behind communications products. Even in computers, however, most flash-memory applications today are replacements for EPROMs and EEPROMs (code storage) rather than mass storage (data storage).
Flash memory may stay as the holder of your BIOS, but the big push for data-storage flash memories probably won't come from computers. Consumer items, such as digital cameras and voice recorders, will play that role, says Alan Niebel, a senior analyst for In-Stat. Luckily for flash-memory makers, demand has outstripped supply, and manufacturers have found a market no matter what their technology. However, Niebel warns, this will start changing later this year, as supplies increase and growth in the computer market slows down. Vendors will feel more pressure in price and technologies. This works for the big companies with advanced technologies, and against both big companies with less advanced technologies and smaller companies that may lack marketing muscle.
New Flash Cards
The prospect of high-volume consumer products, such as digital cameras and voice recorders, is producing yet another form-factor war in the flash-memory industry. The PC Card is too bulky for these applications, so flash-memory makers, camera companies, and other interested parties are touting three new "standards" for tiny solid-state storage. All three of these new cards are about the size of your thumbnail, and all rely on flash memory.
The current leader in the
miniature-card
race is probably SanDisk, with a scaled-down version of its current memory-card technology, called Compact Flash. Since its announcement, about 60 companies have joined the Compact Flash Association (CFA), a consortium backing the SanDisk technology. Among its members are camera companies, such as Polaroid and Eastman Kodak, computer companies, such as Hewlett-Packard, and semiconductor firms, such as Motorola and NEC.
Like a PC Card, Compact Flash includes an A
T-attachment (ATA) controller and memory in the card itself. CFA's competitors call this a drawback, because it increases the cost of the card. CFA's backers call it a feature, because it makes the cards compatible with any computer or other device that uses the ATA disk standard, including most PCs.
Intel is pushing a rather different solution, using an elastomeric connector called Miniature Card (née Mini-Card, but changed for trademark reasons). It is slightly larger than the Compact Flash card. The elastomeric connector has 40 pins (pads, actually) embedded in a rubberized connector, which makes the connector more forgiving at very small sizes than the normal pin-and-socket design. Franklin has been using them for years on its electronic books.
The Miniature Card's strong point is price. Intel and other members of the Miniature Card consortium (including AMD, Fujitsu, and Sharp) point out that the Miniature Cards are cheaper; they contain nothing but flash memory, rather than includin
g a controller chip in each card. Backers of the other approaches point out that this makes them incompatible with PCs and other equipment without special drivers.
Finally, Toshiba is assembling its own group around a technology dubbed the Solid-State Floppy Disk Card (SSFDC). This is a kind of hybrid. The SSFDC can mount in an adapter that fits in a conventional PC Card slot. The adapter contains an ATA controller so that the card-and-adapter combination looks like an ATA hard drive to the computer. However, unlike SanDisk's Compact Flash card, the SSFDC does not need a separate controller chip for each memory card.
At this time, it is far too early to tell which of these three standards is going to succeed. All have important supporters and good technology. Still, it's a good bet that one of them will be very successful indeed.
Flash and Blood
So, with all this support and all these alternatives, why hasn't flash memory toppled the hard drive yet? Price. Flash memory i
s still 10 to 30 times as expensive as hard drive storage. Flash memory's backers point out this differential is much less than it was a couple of years ago. Still, even the most optimistic say it will take perhaps a decade for flash memory to fall to hard drive prices. The less optimistic don't expect that flash memory will ever be as cheap as hard drive storage.
Besides, hard drives are a moving target. Progress in hard drives has been relentless, driving speed and capacity up and costs down.
What about notebooks? Mobile computing seems the ideal market for mass storage not affected by bumps and jolts. But gone are the days when 40 MB was generous for a notebook. Today, the tendency is to load full-featured applications, like Microsoft Word, on notebooks as well as desktop computers. Niebel points out that although 80 MB is an ideal size for a flash-memory "disk drive," effectively no market exists for 80-MB hard drives in PCs, notebooks, or workstations.
PDAs, like Apple's Newton, are
also an ideal application for flash memory. The problem is that they have not taken off yet--and maybe never will. So flash memory isn't getting the boost from them that its backers had hoped.
Choosing which type of memory is best in a real-world application depends very much on the application (see the sidebar
"How Flash Memory Works"
). The ideal of high memory density and rapid random access is as elusive in flash memory as anywhere else. In applications where memory density is most important, slower access may be worth the sacrifice. However, in time-intensive applications, the speed of access may outweigh all other considerations. Consumer applications are typically forgiving on the speed issue, while computer applications usually demand speeds comparable to or faster than a hard drive.
Pressing On to the Goal
Despite the current market reality, most flash-memory companies have split personalities on code storage (cameras and answering machines) and data storage (h
ard drive replacement). They pragmatically recognize where the dough is today and that they need to serve that market to make some.
However, they are also fixated on flash memory's potential for mass storage, and they yearn to sell tens or hundreds of megabytes into a single piece of equipment, rather than a megabit here and a megabit there for code storage. That is why, to capitalize on the coming market, most flash companies are actively promoting mass-storage applications by doing everything from selling PC Card memories to developing new technologies that better fit flash memory to the needs of computer systems.
Flash memory's future is bright, as it continues to move into new market areas. Vendors are making money and developing new technologies. And there's still a flash of hope for competing with hard drives.
Where to Find
AMD
Sunnyvale, CA
Phone: (800) 222-9322
Intel
Santa
Clara, CA
Phone: (408) 987-8080
National Semiconductor
Santa Clara, CA
Phone: (800) 272-9959
SanDisk
San Jose, CA
Phone: (408) 562-0500
HotBYTEs
- information on products covered or advertised in BYTE
photo_link (48 Kbytes)
Flash-memory cards from Toshiba, National Semiconductor, Intel, or SanDisk can give up to 85 MB of solid-state storage--at $20 per MB.
Rick Cook is a freelance writer living in Phoenix, Arizona, who specializes in computers and technology. He is also the
author of several fantasy novels that are full of bad computer jokes. You can contact him on the Internet at
rcook@bix.com
.
The Way of All Flash
